|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
71.194.185.32
About a week ago I published a second installment on modding class D/T amps. In case you want to find the previous two posts – here are the links.http://www.audioasylum.com/forums/amp/messages/111691.html
http://www.audioasylum.com/forums/amp/messages/112104.htmlThis week I’m publishing the third and final installment focusing on power supplies. And remember that improper power supply modding can shorten the lives of amps and humans – so be careful.
Power Supply Mods
One of the big differences between the original Evo 200.2 and the Evo 2 was a much larger power supply – and size matters with class D amps just as it does with class A/B amps. The Evo 2’s 1500 VA toroidal transformer has three identical output windings and is probably the same transformer used in the Evo 6 – except its serving just two channels in the EVO 2.
In the Evo 4 and Evo 6, the rectifiers and caps are located on the Amp Boards. The Evo 2 Amp Board is identical to those in the 4 and 6, but the caps on the Amp Board are augmented by four more capacitors on a separate Power Board and the rectifier on the Amp Board is replaced by three parallel rectifiers on the Power Board. The extra power supply capacity gives the Evo 2 a significant edge over its multi-channel brethren.
The rectifiers employ Diodes Inc. 6A fast recovery diodes and don’t need upgrading. The main power supply caps are Sanwha’s HC general purpose caps – nothing special. I replaced the eight 15,000 uF, 50V caps on the Power Board and Amp Board with Panasonic TSHA caps of the same size and voltage rating. The Panasonic caps are a big improvement providing better dynamics, better separation of instruments and voices, better bass weight – but some improvement at all frequencies. The main power supply caps on the Amp Board are also bypassed with 100 uF Panasonic FC caps near the MOSFETS. These are o.k., but Rubycon ZL caps are a better choice – and because of their small form factor the size can be increased to 470 uF at 50V. Replacing the bypass caps makes additional improvements similar to those from replacing the main power supply caps.
Now if the Evo were really a higher end product – there would be a separate transformer or transformer tap to serve the low voltage circuits on the Amp Board. However, the +/-15V supplies to serve the input op amps are implemented somewhat inelegantly using a voltage divider consisting of two 750 ohm power resistors to drop the voltage of the main supply low enough for the regulators. The regulators are an LM7915CT and an LM340T15. The input caps are 1,000 uF, 35V Panasonic M Series caps – the lowest grade – and the output caps are 220 uF, 25V Nichicon KZ caps – Nichicon’s top audio cap.
To upgrade these supplies, I replaced the cheap, inductive power resistors (R16, R17) with Mills non-inductive power resistors – which make quite a difference in both dynamics and improved treble performance. I also replaced the input caps (C22, C27) with better grade Panasonic FC caps – which is a rather minor upgrade. Later I replaced the Nichicon KZ 220 uF, 25V output capacitors (C37, C38), first with identically rated Elna Silmic II capacitors and then with 470 uF, 50V Rubycon ZL caps. The Silmics were an improvement – but the larger Rubycon caps are a huge improvement.
The +5V power supply serving the Tripath chip is implemented in a manner similar to the +/- 15V supplies, except a 30V Zener diode (D10) is used in conjunction with a 470 ohm power resistor (R36) to get the input voltage low enough for the JRC 7508A regulator. The output cap, located near the Tripath chip (C33), is a 220 uF, 25V Nichicon KZ.
Since the Tripath chip is a high-speed chip – it can benefit from a very clean, very stable power supply, just like the digital sections of CD/SACD/DVD-A players. So, I replaced the +5V volt regulator with an Audiocom Invisus +5V regulator. While this is normally a fairly simple job – there is a complication here. The maximum input voltage for the Invisus is 18V – but 30V is being fed to the regular. So in addition to removing the regulator, the zener diode has to be changed. I used a 15V, 3W DO41 replacement and upgraded the power resistor and input capacitor. For the output capacitor, I used a 270 uF, 16V Sanyo OSCON. Replacing the regulator is a major upgrade – but it should be given the price of the Invisus. The sound becomes richer and fuller. The treble becomes smoother – but there is also more detail. Dynamics pick up after break-in.
There is also a 10V power supply servicing the MOSFET drive portion of the Tripath chip. It must be referenced to the negative power supply rail – and Bel Canto (and Tripath) does this with an LM2594 switching DC/DC converter. Input cap is 33 uF, 50V Nicomp NRSZ cap designed for switching power supplies (C23) and the output cap is a 100 uF, 25V Panasonic FC (C24). These caps are decent, so I left them alone. Three additional caps are located closer to points of use and they are 47 uF, 25V Panasonic types M and SU (C5, C7, C56). I replaced these caps with Sanyo OSCONs 47 uF, 16V caps. The Sanyo caps give a smoother, more natural flow to the music and improved treble performance.
And there’s another thing to do if you’re feeling ambitious. The performance of the transformer can be improved a bit by replacing the ¼” x 20 x 4” magnetic steel bolt that fastens it to the case. This mod has been around for years – but it usually doesn’t show up in commercial products since brass or bronze bolts might not survive shipping and titanium bolts are a little pricy. Some manufacturers use stainless steel, however, which is at least much less magnetic than regular steel.
Finding a brass or bronze bolt is the biggest challenge – especially since many marine suppliers sell mostly stainless steel fasteners these days. I purchased a ¼” x 20 x 4 ½” silicon bronze bolt and flat washers from Jamestown Distributors. And there is another challenge – the nut is secured in place with Loctite Red – which requires the heat of a powerful soldering gun to dislodge. But the results are worthwhile – though subtle. With the bronze bolt in place, the music flows more naturally, sounding less electronic -- particularly in the treble.
And that’s it for now. For about $800 in parts, my Evo 2 Gen II has been transformed from a good amplifier with promise to a great amplifier. And I’m now excited about the future of Class D/T amplification.
Here’s the parts list for this set of mods:
1 ea Sanyo OSCON, 16SP270M $5.76 B&D Enterprises
3 ea Sanyo OSCON, 16SP47M $5.55 B&D Enterprises
1 ea Invisus Regulator, Pos 5V $185.00 Reference Audio Mods
1 ea 1/4x20x4-1/2 Bronze Hex Cap Screw $4.08 Jamestown Distributors
3 ea Bronze Flat Washer $0.21 Jamestown Distributors
1 ea 1/4x20 Bronze Hex Nut $0.08 Jamestown Distributors
3 ea Panasonic FC 1000 uF, 35V Capacitor $3.33 Digikey
8 ea Panasonic TSHA 15,000 uF, 50V $65.36 Digikey
1 ea Zener Diode DO41 3W, 15V $0.30 Digikey
2 ea Mills MRA-12 750 ohm Resistor $7.90 PartsConnexion
1 ea Mills MRA-12 470 ohm Resistor $3.95 PartsConnexion
4 ea Rubycon ZL 470 uF, 50V Capacitor $12.00 Reference Audio Mods
Total $293.52
Follow Ups:
Thanks, again for such a comprehensive description of what you have done . The results that you heard at various junctures is what I find most informative.If you will at after some time report if the sound change any as the mods burn in.
DaveT
Thanks.I finished the last of the mods two months ago and everything is pretty well broken in. However, I’m thinking about a couple more mods and if they work out, I’ll post again.
Thanks for the clarification on the sonic observations and break in time. I look forward to reading about the next series of mods.
*
This post is made possible by the generous support of people like you and our sponsors: